XAI-FinCrime
Sep 1, 2025
·
1 min read
Funding: Innosuisse · September 2025 – August 2028
The project aims to transform financial crime prevention by leveraging machine learning, generative AI, and explainable AI (XAI) to enhance fraud detection and compliance management. It uses ML models for transactional behavior analysis and multimodal LLMs for unstructured data processing, with counterfactual explainers to improve AI decision transparency.
Objectives
- Significantly reduce the typical False Positive rate of 95% found in current rule-based industry standards
- Improve F1-scores, accuracy, and True Positive rates compared to existing solutions
- Enhance user understanding of AI systems to reduce investigation effort per alert
- Validate innovations through pilot studies with financial institutions
Authors
Professor
Silvia Santini is an Associate Professor at the Faculty of Informatics of USI since September 2016, where she co-leads the People-Centered Computing Lab together with Prof. Marc Langheinrich. From July 2014 until August 2016 she held an Associate Professor position at TU Dresden, where she led the Embedded Systems Lab. From October 2011 until July 2014 she was an Assistant Professor at the Department of Electrical Engineering and Information Technology of TU Darmstadt, Germany, where she led the Wireless Sensor Networks Lab. From 2009 until 2011 she was a postdoctoral researcher in Prof. Friedemann Mattern’s Distributed Systems Group at ETH Zurich, and from November 2010 until February 2011 she joined Leonidas Guibas’s research group at Stanford University as a visiting scholar. Silvia completed her PhD under the supervision of Prof. Friedemann Mattern at ETH Zurich in 2009, and graduated in Telecommunication Engineering (with honors) from the Sapienza University of Rome, Italy, in May 2004.